A comparison you can recompute.
How do ten prespecified transcripts compare descriptively between F. nucleatum-exposed and medium-control HT-29 samples in GSE90944?
What changes in
this selected view?
The original investigators exposed HT-29 colorectal cancer cells to F. nucleatum and submitted processed RNA-seq gene-count tables. Predome re-reads those tables, checks the six complete library denominators and computes a small, source-linked descriptive comparison.
This case asks about the displayed transcript counts. It does not test chemotherapy response, patient outcomes, protein function or a causal mechanism.
Inspect the official GEO recordWatch the checks.
Read the limits.
A 25-second visualization of the saved calculation outputs. The source checks, sample values and result fingerprint can all be inspected below.
Keep the individual points.
Dots: submitted samples. Dashed lines: arithmetic means. No confidence interval or significance test is shown.
- Control mean
- 55.02
- Exposed mean
- 76.60
- Ratio of group means
- 1.392×
- Log₂ ratio
- 0.477
CPM = count ÷ all assigned gene counts × 1,000,000. The ratio uses the two arithmetic group means, without a pseudocount.
| Sample | Group | Source count | Library denominator | CPM |
|---|---|---|---|---|
| NC1GSM2417930 | Medium control | 1,022 | 18,618,711 | 54.891 |
| NC2GSM2417931 | Medium control | 949 | 17,041,899 | 55.686 |
| NC3GSM2417932 | Medium control | 887 | 16,278,512 | 54.489 |
| Fn1GSM2417927 | F. nucleatum | 918 | 15,330,313 | 59.881 |
| Fn2GSM2417928 | F. nucleatum | 1,488 | 18,457,238 | 80.619 |
| Fn3GSM2417929 | F. nucleatum | 1,396 | 15,633,744 | 89.294 |
Inputs. Checks. Outputs.
These are actual stages of the published Python run. They use deterministic code; no language model was called.
Source verification
- Started
- 2026-11-02 06:00:05 UTC
- Elapsed
- 147.035 ms
{
"archive_sha256": "e4c78d89055b94cf18730e098aa8326ad5a7ae8468b0bb3392102d4085eb1459",
"tables": 6,
"table_rows": {
"GSM2417927": 56643,
"GSM2417928": 56643,
"GSM2417929": 56643,
"GSM2417930": 56643,
"GSM2417931": 56643,
"GSM2417932": 56643
},
"source_date": "2017",
"source_kind": "submitter-provided processed HTSeq counts; not FASTQ"
}A descriptive observation.
A bounded conclusion.
Transcript abundance differences in this selected public-data subset are descriptive. They do not establish autophagic activity, a causal mechanism, chemotherapy resistance or clinical efficacy.
The ten genes were chosen for display before this computation. They are not a discovered signature or an unbiased genome-wide screen.
Alternative explanations and missing evidence
- Independent biological replication is not established by the submitted metadata.
- Library-size adjustment does not remove batch or composition effects.
- RNA abundance is not a direct measure of protein activity or autophagic flux.
- The displayed subset contains no chemotherapy arm, viability assay or patient response.
Proposed next verification · not executed
- Review sample independence and batch metadata.
- Repeat a prespecified analysis on an independent dataset.
- Use an appropriate experimental assay before claiming protein function or drug response.
Take the record with you.
Download the original processed-count archive, source metadata and an offline reproduction script. Timestamps will change on a new run; the calculation-result fingerprint should remain the same.
- Original archive SHA-256
- e4c78d89055b94cf…2d4085eb1459
- Calculation result SHA-256
- 2275357f2ef27fcb…7b2c12852ee1
A file fingerprint checks content consistency. It is not independent proof of creation time or scientific truth.
Bring a question.
Define a useful next check.
Start with a bounded computational or forecasting task and a clear route to verification.
Explore collaboration process